Problem with binomial logistic regression

[drfuad]

Hi,
I am a new user of jamovi and loving the simplicity. I know nothing about coding. During an analysis using a dataset having 382 cases, I stumbled on a problem while performing binomial logistic regression. I entered my DV and 3 IVs as the factors. The output generates the folloing massage:

Table$setRow(): value 'est' is not atomic.

Capture.JPG (45.2 KiB) Viewed 2632 times

I was wondering, am I doing something wrong? How can I fix this?
I am also attaching the dataset containing those variables.
Thanks,

[MAgojam]

I am a new user of jamovi and loving the simplicity. I know nothing about coding.

Hey @idrfuad,
a quick and easy answer.

The problem that has arisen is probably due to how you have coded two of your variables.
In the coding you added the level "CP not having" for the variable rec_owningageCP and the level "Non-use" for the variable Rec_use_years, which should be treated as missing data.
If so, the BLR analysis resolves, as you can see by taking a look at the screenshot.

screenshot_BLR.png (279.58 KiB) Viewed 2624 times

Cheers,
Maurizio

[drfuad]

Thank you for the swift response. It does work when I follow your instruction. However, Non-use and cp not having is the reference level needed for the analysis. Is there any other way to keep them in the model and correct the error massage?
Thanks in advance.

[MAgojam]

Hey @drfuad,
to continue looking at your dichotomous outcome with a binary logistic regression is to find the probability that there are sleep disturbances in cell phone users you would have to free the 180 cases from the "CP not having" and "Non-use" level metch that I considered missing because it's not clear to me how you got there.
Look at the screenshot (attached) of your cases, where I drew attention only to some of the 180 cases that I presented for the variable Rec_CP_use an hourly class of use of the CP, while the other two variables report "CP not having" and "Not -use".

levels.png (131.47 KiB) Viewed 2599 times

What does it mean that the subject does not have a personal cell phone and does not use it, but still uses a cell phone that is not hers for a certain number of hours per day?
How did you enter the data from which you started?
Maybe something went wrong with the recoding?

Cheers,
Maurizio

[Ravi]

So in terms of what's going wrong in the model is that it seems that the data you are using in the model are highly collinear resulting in a singular fit. You can see in the data that the `CP not having` level of `rec_owningageCP` and the `Non-use` level of `Rec_use_years` only occur together (and not with any of the other levels) and therefore are highly dependent on each other.

I added a more informative warning message to the analysis so that at least it's a little clearer what's going on here, see https://github.com/jamovi/jmv/pull/384.